v7.8.0

Minor release headlined by agent reasoning state support and multi-LLM configuration in the BAF (BESSER Agent Framework) code generator. It adds new agent metamodel classes (ReasoningState, Tool, Skill, Workspace) and multi-LLM factory methods on Agent (new_llm, set_default_llm, new_tool, new_skill, new_workspace, new_reasoning_state), surfaced through the BAF generator template, the agent-model code builder, and the JSON↔B-UML converters. Compatible with BESSER-Agentic-Framework v4.4.0.

Agent Reasoning State and Multi-LLM Support

BAF Generator — multi-LLM handling

• Introduced a unified default_llm variable in generated agent code, replacing the previous reply_llm approach. It is set from the agent configuration or the first available LLM; if none is configured it is None, so a clear AttributeError is raised at runtime rather than a silent no-op.
• All LLM invocations (DBReply, LLMReply, intent-classifier, fallback actions) now resolve the LLM by name at runtime, falling back to default_llm — making it easy to run different actions against different models in one agent.
• Intent-classifier LLM selection uses a prioritised search across the per-classifier config and agent-level defaults.

RAG improvements

• RAG configurations resolve their LLM at runtime via llm_name, falling back to default_llm. The separate rag_llm variable is no longer emitted.

Tools, Skills, and Workspaces

• Generated agent initialisation now registers agent tools, skills, and workspaces, completing the round-trip from B-UML agent model to runnable BAF code (compliant with the BAF v4.4.0 new_tool/new_skill/new_workspace signatures).

Reasoning State support

• Code generation for ReasoningState objects: the generator emits a new_reasoning_state(...) call with the appropriate LLM reference (falling back to default_llm) and configuration (max_steps, enable_task_planning, stream_steps, system_prompt, fallback_message).

Editor (frontend)

• Agent editor gains the reasoning extension UI (AgentLLM, AgentTool, AgentSkill, AgentWorkspace, AgentReasoningState), a multi-LLM configuration panel, per-type colors/icons/distinct shapes, NN-style titled palette sections, and a sidebar that defaults to expanded.